Research On Image Enhancement Of Cone Beam CT

As an advanced imaging technique, Cone-beam CT(CBCT) has the advantage of low dosage, large field of view and high-spatial resolution, which has been widely used in many field, such as medical diagnosis and treatment, industrial X-ray dectection. However, the quality of reconstructed image is subject to the manufacture technology and assembling process, leading to different types of artifacts which hamper the evaluation of the structures of object. Three kind of degraded factors are studied in our work, including defective projection data, geometric artifact and ring artifact.In order to improve the quality of slices, the degradation mechanism of CBCT is analyzed, the studies are based on the spatial domain and statistical characteristics of projections, sinograms, slices and polar coordinate transformation images. On the one hand, as for stripes and bad pixels exist in each projection, the defective map of flat panel detector and histogram are used to locate the position of pixels, which are inpainted by eight-neighbor nearest interpolation. The light and dark field projections are introduced to correct the nonuniformity. Besides, a reference block correcting method is used to ensure reconstructing high-quality slices. On the other hand, taking the geometric artifact into account, the calibrating phantom was scaned angularly in 360°. After contrast enhancement, segmentation, point-like objects central coordinate location and the trajectory reconstruction, then geometric parameters are calculated. Futher, to the structural ring artifact in slices, we firstly compared serveral typical sinogram based pre-correction and coordinate tranformation based post-correction methods. In consideration of their pros and cons, a variation-based ring artifact correction method with sparse constraint is proposed by fully utilizing the directivity and sparsity of stripes in polar image. The alternating direction method of multipliers(ADMM) is introduced to split the hard to solve minimization problem into serveral easy to solve subproblems.Experimental results on simulated and real data demonstrate that the degradation of slices in the process of reconstruction is avoided because of the effective improvment of projections quality and precise calibration of geometric parameters. The proposed variation-based ring artifact correction method has the ability to correct artifact as well as preserve the structures and details, to futher improve the image quality. The studies of our work should be a support of subsequent reseaches on CBCT.